News & Events

Firearm assaults too close to home?

UW EPI NEWS | May 3, 2019
3 minutes to read

When it comes to firearm assaults, the neighborhoods where people live are not necessarily where they are shot. A new study conducted by researchers at the University of Washington School of Public Health found that nearly three-quarters of firearm injuries and deaths in King County occurred beyond the immediate area surrounding the victim’s home.

Previous research has found that assaults involving a firearm are clustered in ‘hotspots’ within particular neighborhoods or cities, which has implications for the individuals who live in those areas. Using the same methodology, Brianna Mills, a doctoral graduate from the Department of Epidemiology and a research scientist at the Harborview Injury Prevention & Research Center, found that more injuries happened in certain neighborhoods to people who don’t live there than to the residents of that area. When residential addresses are used as proxies for injury location regardless of how geographically accurate they may be, it introduces bias to the data, possibly underestimating the number of firearm assaults in those areas.

Published in the journal Injury Prevention, this brief report examines the potential degree of misclassification resulting from using residence location as a stand-in for injury location, and how missing location data is associated with other known factors, such as homelessness.

Mills used data from the Harborview Medical Center Trauma Registry and Washington State death records to analyze 670 fatal and non-fatal firearm assault injuries that occurred in King County between 2010 and 2014. She found that the median distance between the location of injury and the individual’s address was almost 4 miles, with the greatest distance between residence and injury location seen amongst young people aged 18–34.

Mills also discovered most missing residential addresses in these records were due to assaults on the homeless or people who live outside the county. This type of missing information introduces bias into a residential-based database because it excludes the homeless and unstably housed or it misrepresents an injury hotspot where injuries occur primarily to nonresidents.

The misclassification of injury location focused on a particular type of assault may impact the effectiveness of location-based interventions designed to reduce the risk of firearm injury. Mills says researchers should consider when residence location and injury location should be treated as geographically distinct.

This study was part of Mills’ doctoral dissertation while at the University of Washington Department of Epidemiology. It was supported by the Center for Studies in Demography and Ecology Shanahan Fellowship and the Harborview Injury Prevention & Research Center.